Fast response measurements of particle size distributions, bulk submicron particle composition, and single particle composition were made aboard the NOAA WP-3D research aircraft in the free troposphere over the eastern Pacific Ocean and the western coast of North America. Measurements of gas-phase compounds and meteorological analysis show evidence of long-range transport of layers of aerosol particles from anthropogenic and biomass-burning sources in eastern Asia. Layers of crustal particles with no evident sources were also encountered. Measurements of substantially enhanced particulate sulfate mass and gas-phase H2SO4 encountered during one transport event were interpreted with the aid of a numerical model of particle formation and growth and trajectory analysis. The observed particle size distributions and sulfate mass concentration were consistent with the simulation, indicating that the particles were formed over the mid-Pacific from gas-to-particle conversion following long-range transport of SO2 through a midlatitude cyclonic system. Such cloud systems appear to effectively scavenge most pre-existing particle mass, but can allow transport of gas-phase precursors which substantially alter downstream particle microphysical and chemical properties.
 Several studies indicate that Eurasian sources of trace gases and particles, including urban and industrial emissions, biomass combustion, and dust from desert regions, can reach the troposphere over the North American continent [Jacob et al., 1999; Jaffe et al., 2003, 1999; Kasischke and Bruhwiler, 2002; Price et al., 2003]. Emissions from regions of China and southeast Asia are a particular concern because of the potential for growth following increasing industrialization and urbanization [Streets and Waldhoff, 2000]. The potential effects of these emissions include: (1) direct changes to the earth's radiation budget due to increased optical depth and increased light absorption; (2) indirect radiative effects due to changes in the cloud nucleating ability of the particles; (3) changes in the gas-phase chemistry downwind of the sources which might have large-scale effects on OH concentrations and the oxidizing capacity of the remote troposphere; and (4) increases in the background concentrations of controlled pollutants and their precursors, which might cause regulatory exceedances and increased health impacts when North American sources are added to an elevated background [Berntsen et al., 1999; Fiore et al., 2002].
 Both Asian source regions and North American receptor regions have been studied with the aim of characterizing trans-Pacific transport of pollutants. Recent studies have used in situ measurements, remote sensing observations, and chemical transport models to study the emissions and transport of gases, particles, and particle precursors from Asia over the western Pacific Ocean. These studies have focused on episodes of trans-Pacific transport of soil dust [Husar et al., 2001] and forest fire smoke [Kasischke and Bruhwiler, 2002], and on the characteristics of Asian urban and industrial pollution detected east of the Asian coastline [Bahreini et al., 2003; Miyazaki et al., 2003; Clarke et al., 2004]. A few observations have been reported of transport of anthropogenic pollutants from Asia to the eastern Pacific or the western coast of the North American continent. These studies [Jaffe et al., 2003, 1999, 2001; Price et al., 2003] report a wide variation in concentrations of CO, O3, ethane, propane, and particle light scattering in layers of pollutants transported from Asia. In most of these layers, light scattering due to particles was significantly enhanced above median conditions. Soil dust particles associated with enhanced mixing ratios of CO and the anthropogenic markers ethane and propane were sometimes present. Other studies [Clarke et al., 2001; Moore et al., 2003] describe the characteristics of apparently similar layers containing particles of both crustal and anthropogenic origin encountered in the free troposphere between Hawaii and North America.
 Emissions from Asian sources may produce more than episodic effects on tropospheric composition. Several modeling studies [Fiore et al., 2002; Jacob et al., 1999; Li et al., 2002] and observations [Jaffe et al., 1999, 2001; Parrish et al., 1992] indicate that transoceanic transport of anthropogenic emissions may produce a substantial perturbation to background gas-phase chemistry, including increases in ground-level O3 mixing ratios. A significant contribution to background aerosol properties due to soil particles from Asian sources has been reported at high-altitude surface stations in the western United States [vanCuren and Cahill, 2002; vanCuren, 2003].
 From these studies, it is evident that several sources, principally soil dust, biomass burning (both from forest and agricultural fires and from biofuel combustion), and urban and industrial emissions, influence the properties of aerosols advected in the free troposphere to the western coast of North America. The relative importance of intercontinental and North American aerosol sources on North American tropospheric aerosol properties must ultimately be evaluated in large part by using carefully constructed and constrained global chemical transport models. By evaluating the detailed particle microphysics and composition and the gas-phase environment and meteorology associated with aerosols transported from Eurasia over the western coast of North America, an improved understanding of aerosol source characteristics and the transport and transformation processes can be developed to help constrain such model simulations.
 In this work, data from fast-response, airborne gas- and particulate-phase sensors are used to evaluate the chemical and microphysical properties of aerosols approaching the western coast of North America. This work is unique because it examines in detail the chemistry and microphysics of the aerosol particles while using the simultaneously measured gas-phase measurements of reactive and tracer species to identify different sources and to evaluate processes affecting aerosol properties during transport. These measurements of particulate and gas-phase pollutants and tracer species were made from 22 April to 19 May 2002 on the National Oceanic and Atmospheric Administration (NOAA) WP-3D aircraft as part of the Intercontinental Transport and Chemical Transformations (ITCT) program. Two of the primary goals of ITCT are to improve quantitative understanding of the intercontinental transport of ozone, particles, and their precursors, and to quantify the effects of this transport on climate and regional air quality. The airborne measurements were complemented by ground-based sampling, meteorological analysis, and numerical modeling of the large-scale transport and chemical evolution of pollutants from Northern Hemisphere sources.
 This work describes and discusses observations of particle size distributions and composition obtained in the free troposphere in air not directly affected by North American sources. The statistics of particle characteristics in this environment are presented, and two cases of significant transport from Asia to the free troposphere over the western coast of North America are evaluated in detail. For these two cases, the measurements help identify likely sources of the particles and the processes affecting particle properties during transport. Separate reports more completely describe the characteristics, sources, transport and distribution of reactive gas-phase compounds [Nowak et al., 2004], and acetonitrile and other trace species encountered during ITCT [de Gouw et al., 2004; Nowak et al., 2004]. Cooper et al.  provide a detailed meteorological analysis of one of the two transport events described here.
2. Instruments and Methods
 The NOAA WP-3D, a large marine reconnaissance aircraft (http://www.aoc.noaa.gov), was operated from Monterey, California, during April and May 2002. To study long-range transport, the aircraft made 13 flights of ∼7 hr duration that included long horizontal transects, stairstep climbs, and vertical profiles over altitudes from 0.1 to 7.9 km. The flight patterns were designed to intercept transported plumes of Eurasian emissions predicted by tracer transport models [Cooper et al., 2004; Forster et al., 2004], and to study the effects of local emissions from ships, biomass fires, and North American cities [de Gouw et al., 2004, 2003b; Neuman et al., 2003].
 The aircraft was equipped with instruments for measuring a large variety of gas-phase compounds and particle composition and size distribution. Many of the instruments have response times of 1 s, allowing detailed characterization of the horizontal and vertical distribution of the measured parameters. The measurements of primary importance to this analysis are described in this section. Throughout this work, the term “aerosol” is used to mean the measured mixture of gas-phase and condensed-phase components while “particle” refers only to the condensed phase component of the aerosol.
2.1. Particle Size Distribution Instruments
 Particle size distributions from 0.004–7 μm diameter (Dp) were determined with 1 s time resolution by combining measurements from a 5-channel condensation particle counter (CPC) [Brock et al., 2000], a modified LasAir 1001A laser optical particle counter (OPC, Particle Measuring Systems, Boulder, CO), and a white light OPC (a modified model 208A, Climet Instruments, Redlands, CA). Before, during, and after the ITCT 2002 field project, the three particle sizing instruments were calibrated with nearly monodisperse (NH4)2SO4 particles with Dp from 0.05 to 1.6 μm [Brock et al., 2003]. Prior to the mission the instruments were calibrated with particles of varying compositions and diameters ranging from 0.004 to 10 μm [Brock et al., 2002]. Particles were dried by heating the sample airstreams to ∼35°C before they were measured; relative humidities, measured at the point of particle detection for the white light OPC, were <18% for all data reported here.
 The two optical particle counters sampled from a low turbulence inlet (LTI) that used active boundary layer suction to virtually eliminate turbulence [Brock et al., 2003; Wilson et al., 2004]. The LTI was mounted on the forwardmost left window immediately behind the cockpit, well forward of upwash flow around the wing root area. Corrections to the measured size distributions to account for inertial enhancements of large particles in the LTI were calculated using computational fluid dynamics modeling [Brock et al., 2003; Wilson et al., 2004] and were applied to the data. Conductive rubber tubing carried the sample airstream 5.2 m to the OPCs. Losses of particles due to gravitational and inertial deposition, which were significant for particles with Dp > 3 μm, were estimated using empirically based methods [Baron and Willeke, 2001] and corrections were applied to the data.
 A Monte Carlo simulation was used to propagate the random uncertainties due to calibration of particle size and concentration, and to in-flight measurements of flow rate, temperature, and pressure, through the numerical data processing algorithms, using typical conditions measured in the free troposphere during ITCT 2002. The resulting estimated measurement precisions are given in Table 1. Accuracies were estimated by calibrating the instruments with particles of differing composition and refractive index [Brock et al., 2000], and propagating the range of responses through the data processing algorithms to the final products (Table 1). All particle parameters are reported in quantity per cubic centimeter of air at standard temperature and pressure (STP, 1013 hPa and 273.16 K.)
Table 1. Uncertainties for Particle Number (N), Surface Area (S) and Volume (V) Concentrations and Submicron Particle Compositiona
Particle diameter ranges are 0.004–7.0 μm, 0.15–1.0 μm, and 1.0–7.0 μm. Uncertainties in N (particle number), S (surface area), and V (volume) are derived from Monte Carlo simulations of the propagation of calibration and concentration errors.
NA, not applicable. Detection limit is highly variable depending on particle concentration and size.
Precision is highly variable depending on whether sub- or super-micron particles dominate volume.
NC, not calculated. Uncertainty is dominated by calibration precision.
 A Particle-Into-Liquid Sampler, or PILS [Orsini et al., 2003; Weber et al., 2001], followed by ion chromatographs (IC), was used to measure the bulk concentrations of submicron (Dp < 1 μm), water-soluble, particulate inorganic ionic components. This system injects saturated water vapor into a sample airstream, condenses water on the particles, and impacts the resulting droplets onto a plate flushed with an aqueous flow containing a Li+ dilution marker. The resulting liquid stream is then analyzed by two ion chromatographs, one for anions and one for cations. During the ITCT 2002 project, 2.7-minute integrated samples were obtained every four minutes for both anions and cations.
 The PILS system operated downstream of the LTI and used an impactor to remove particles with dry aerodynamic Dp > 1.0 μm from the sample airstream, while denuders between the impactor and the saturator removed gas-phase SO2, NH3, and HNO3 prior to measurement. Estimated accuracy and precision for the PILS ion chromatography are given in Table 1. Transmission losses of particles with Dp < 1.0 μm in the laminar flow within the sampling lines were calculated to be negligible [Baron and Willeke, 2001]. The PILS data are in units of μg m−3 and have been corrected to STP. Poor charge balance and a slow time response to airmass changes indicate that the ammonium measurement in air with near-background concentrations was affected by contamination. Therefore the ammonium data are not used in this analysis, although the signal is judged to be sufficient for qualitative use at the higher concentrations encountered during flights affected by local urban and agricultural emissions [Neuman et al., 2003].
2.3. Single Particle Mass Spectrometer
 The Particle Analysis by Laser Mass Spectrometry, or PALMS, instrument [Thomson et al., 2000] was used during ITCT to obtain real-time information on the composition of individual particles. The PALMS instrument vaporizes and ionizes single particles with an excimer UV laser, then measures the mass/charge ratio of the ions using a time-of-flight mass spectrometer. Owing to varying instrumental sensitivity caused by interactions between the composition and size of the particles and the efficiency of laser ablation and ionization, it is not possible to use the PALMS to make quantitative measurements of the concentration of the chemical constituents within most tropospheric particles. However, because the technique is very sensitive to the trace presence of many particulate components [Middlebrook et al., 1997; Murphy et al., 1998], it can be used to derive the statistics of the occurrence of classes of particles as a fraction of the total analyzed population of particles. The PALMS can be operated in either a positive or negative ion detection mode; in this work, we focus primarily on data from the positive ion mode.
 As configured in ITCT 2002, the PALMS inlet sampled particles with Dp between 0.3 and 3.0 μm with >50% efficiency, while particles with Dp as small as 0.2 μm and as large as 5.0 μm were sampled with lower efficiency. Positive mass spectra of the ions were placed into groups based on the mass peaks observed using a clustering algorithm [Murphy et al., 2003]. These clustered groups were combined into 4 non-overlapping categories based on similarities in analyzed ion composition; >80% of all the analyzed particles were encompassed by the four categories. The four categories of positive mass spectra are: (1) containing only carbonaceous (non-soot) and sulfate peaks, which are typical of free tropospheric particles [Murphy et al., 1998], (2) containing only carbonaceous and sulfate peaks and NO+, which may be produced by any nitrogen-containing compound in the particle, but most commonly represents the presence of nitrate or ammonium; (3) containing potassium and carbonaceous matter, often with NO+ and/or sulfate, believed to originate primarily from natural or anthropogenic biomass combustion sources [Hudson et al., 2004]; and (4) containing iron, silicon, and/or aluminum, which are assumed to be crustal in origin.
2.4. Other Measurements
 The analysis presented here makes use of measurements of several photochemically reactive gases and tracer compounds to identify air influenced by fossil fuel or biomass burning, to provide bounds to the time since emission, and to evaluate the mass budget of condensable compounds. The gas-phase species measured with 1 s time resolution include the sum of the reactive nitrogen compounds (NOy), NO, NO2, HNO3, O3, SO2, and CO. The accuracy and precision of each of these measurements, other than O3, were quantified by in-flight calibration by standard addition of calibration gas and by zeroing [Holloway et al., 2000; Neuman et al., 2002; Ryerson et al., 1998, 2001]. The detection limit for the SO2 instrument of 350 pptv was greater than ambient concentrations for most of the data analyzed here. The fast-response chemiluminescent O3 sensor was calibrated in-flight with a slower response, primary standard UV absorption instrument [Ryerson et al., 1998]. Acetone, acetonitrile, methanol, benzene, toluene and several other organic compounds were measured with ∼9 s response time using a proton transfer reaction-mass spectrometer (PTR-MS) [de Gouw et al., 2003a; Neuman et al., 2002; Ryerson et al., 2001]. Gas-phase H2SO4 was measured by chemical ionization mass spectrometry (CIMS) [Eisele and Tanner, 1993] for a 1.1 s period every 4 s with an accuracy of ±35%, a precision of ±1 × 106 molecules cm−3, and a 3-σ detection limit of ∼6 × 106 molecules cm−3. In addition to these real-time measurements, approximately 80 individual whole-air canister samples were collected periodically over 6–18 s intervals during each flight and were subsequently analyzed by chromatographic techniques [Heidt et al., 1989; Schauffler et al., 1999]. Quantified species used in this analysis included alkyl nitrates, alkanes, and alkenes. The NOAA Aircraft Operations Center (http://www.aoc.noaa.gov) operated and maintained instruments for the navigational and meteorological measurements used in this analysis.
2.5. Data Selection
 The ITCT 2002 data set includes measurements made in free tropospheric air unaffected by immediate North American sources of pollution, as well as observations obtained in air influenced by urban, industrial, biomass, and ship emissions directly associated with the North American continent and its coastal waters. To focus on a primary goal of this study—the characteristics and consequences of long-range pollutant transport arriving from upwind trans-Pacific sources—only a subset of the ITCT 2002 data are considered here. These selected data are restricted to periods when the aircraft was in air >1 km above the continental planetary boundary layer (PBL) within 250 km of the coast, or above the marine boundary layer (MBL). The height of the PBL and MBL was subjectively determined on a case-by-case basis by sampling location and altitude and by sharp vertical gradients in temperature, relative humidity, CO and O3 mixing ratios, and particle concentrations. This subjective method was used because the wide range of measured values for these parameters within the free troposphere precluded development of rigorous objective criteria for identifying the MBL/free troposphere boundary. Data were included in the analyzed data set only if the measured winds at the sampling location had a predominant westerly component, and if backward trajectories, which were calculated for one-minute time intervals along the flight [Cooper et al., 2004], showed transport over the Pacific without direct influence from North American sources. Additional data filtering removed encounters with aircraft plumes, which were characterized by high concentrations of particles, NOy, NOx, and CO2, without commensurate CO increases, over short distances. Data obtained in clouds were excluded based on relative humidity measurements and the presence of a detectable liquid water content from a liquid water probe, and/or video from a nose-mounted camera. Since many of the flight plans were designed to sample long-range transport, the filtered data set includes ∼43% of the total data collected during the airborne portion of the ITCT project. The data and analysis presented here use only this selected subset of the aircraft data.
 Enhanced concentrations of trace gases were used to identify direct, long-range transport of pollution. As noted by Nowak et al. , the unambiguous identification of transport events is difficult, since most air parcels in the free troposphere of the Northern Hemisphere are affected to some degree by anthropogenic and biomass burning emissions that have been transported over widely varying time and spatial scales and from diverse sources [Novelli et al., 1998]. On the basis of statistical arguments, mixing ratios of CO exceeding 150 ppbv have been identified as belonging to a population significantly above the expected range of free tropospheric background CO values [Nowak et al., 2004], and are used to identify substantial transport events due to anthropogenic and biomass burning sources directly upwind of North America.
 In the following section, the statistics of occurrence for particle properties during the selected portions of the ITCT 2002 project are quantified. We then examine in detail the chemical, microphysical, and meteorological characteristics and the spatial distributions of pollutant and dust layers found during discrete and identifiable transport events encountered during two flights.
3. Observations and Analysis
3.1. Statistics for All Flights
 Statistics of occurrence for particle physical and chemical parameters were calculated for periods of flight above the planetary or marine boundary layer under Pacific flow conditions as described in section 2.5. Because the aircraft was directed into regions predicted by trajectory models to contain enhanced concentrations of pollutants transported from Asia, these statistics may be biased toward transport events. However, in only 7% of the selected data set did CO exceed the 150 ppbv criterion established for direct pollutant transport. Since significant transport events were relatively infrequent, percentile statistics calculated from the data set should be less sensitive than means and standard deviations to possible biases. Therefore we report and interpret medians and other percentile statistics of the measured particle parameters.
 Measured dry particle size distributions were used to calculate particle number, surface, and volume concentrations in three size classes: 0.004–7.0 μm (total), 0.15–1.0 μm (accumulation mode), and 1.0 to 7.0 μm diameter (coarse mode). These size ranges were selected based on the observed particle modes present and on the detection ranges of the instruments. These calculated parameters were approximately lognormally distributed. The 10–90 percentile range for total particle number, surface, and volume concentrations were 350–1200 cm−3, 9.1–43 μm2cm−3, and 0.37–4.2 μm3cm−3, respectively (Table 2).
Table 2. Percentile Statistics of Occurrence of Particle Number (N), Surface Area (S) and Volume (V) Concentrations and Submicron Particle Composition in Air Above the Planetary Boundary Layer and Not Directly Affected by North American Emissions During the Entire ITCT Programa
Particle diameter ranges are 0.004–7.0 μm, 0.15–1.0 μm, and 1.0–7.0 μm. Statistics for particle volume from 1.0–7.0 μm are from 10 s means; all other integrated size distribution values are from 1 s data. Units are number, cm−3; surface, μm2cm−3; and volume, μm3cm−3. PALMS data were divided into running blocks of 200 consecutively analyzed particles (equivalent to ∼1 minute of flight time); values show fraction of analyzed particles per 200-particle block that fall within one of 4 PALMS mass spectra categories: A, sulfate-carbonaceous (non-soot); B, sulfate-carbonaceous-NO+; C, containing K ions (probably biomass burning); D, containing ions from Fe, Si, and/or Al, presumably crustal. N, particle number; S, surface area; V, volume.
 The PILS-IC concentrations for submicron water-soluble compounds were generally quite low, with median values of SO4= of 0.13 μg m−3 and with many samples at or below the detection limit for the 9 ionic species reported.
 Statistics for the PALMS data (Table 2) were calculated based on moving boxcar sums from 200 analyzed particles. Data from the PALMS instrument show that particles containing carbonaceous material and sulfate (A), carbonaceous material, sulfate (B), and NO+, and potassium (most commonly associated with biomass burning) (C), each contributed about 1/4 of the total analyzed particles. Despite the bias in the PALMS sampling toward relatively large particles with Dp between 0.3 and 3 μm, particles containing crustal material were relatively infrequent over the entire free tropospheric data set (on average <6% of analyzed particles). These statistics are similar to those derived from analytical electron microscopy of particles collected by impactor at 8.6 km altitude over California, showing ∼4% of particles to be crustal in origin [Sheridan et al., 1994]. The remaining analyzed particles did not fall into one of the four categories, and represent a variety of compositions.
 There were several encounters with air showing evidence of anthropogenic and/or crustal sources of particles, resulting in particle properties substantially exceeding the 90th percentile values in Table 2. The following sections present and discuss two of these episodes, with the goal of describing and contrasting the particle characteristics of the different transport events and explaining the observed variability in terms of differing sources and transport processes.
3.2. 5 May 2002
 On 5 May 2002, the WP-3D intercepted air over the Pacific Ocean off the western coast of North America that was predicted by tracer transport models to contain elevated concentrations of CO from Asian anthropogenic and biomass burning sources [Cooper et al., 2004; Forster et al., 2004]. The aircraft encountered layers of air with concentrations of different pollutants and tracers significantly exceeding the range of mixing ratios expected for background air [de Gouw et al., 2004; Nowak et al., 2004]. In this section, we describe the particle characteristics within each of three vertically contiguous layers and evaluate their likely sources and transport mechanisms.
 From 22:40 to 00:30 UTC, when enhanced CO mixing ratios were measured (Figure 1), NOy and total particle number concentrations were positively correlated with CO with linear coefficients of regression (r2) of 0.91 and 0.56, respectively. Many other trace gases, including ethane, propane, butanes, pentanes, acetylene, and CO2, were significantly enhanced within the CO-rich layers, and the ratios of these compounds strongly indicate transport from anthropogenic sources in eastern Asia over a period of 5–7 days [de Gouw et al., 2004; Nowak et al., 2004]. In contrast, submicron particle volume was not strongly correlated with CO, and supermicron particle volume was negatively correlated over portions of the encounter (e.g., from 23:30 to 00:00 UTC). The water-soluble ionic particulate constituents were not significantly enhanced above the median background levels reported in Table 2. Neither H2SO4 nor SO2 showed detectable increases in the layers of enhanced CO.
 Within this region of enhanced CO, the WP-3D executed a stairstep climb from 3.8 to 7.9 km altitude at latitudes between 34 and 36oN after 23:50 UTC (Figures 1 and 2). Particle size distributions measured during this climb were averaged into 10 m vertical bins (approximately 4 s of data per bin) to produce vertical profiles of particle number and volume distributions (Figures 2c and 2d). The concentrations of particles with Dp < 0.2 μm were enhanced in the layer from 5.3 to 6.8 km containing mixing ratios of CO > 200 ppbv. In portions of this layer, number concentrations of particles with Dp < 0.01 μm were elevated, suggesting that new particle formation had recently occurred. Despite the high concentrations of these small particles, neither submicron particle volume (Figure 2e) nor water soluble ionic particulate compounds were enhanced above background within this layer.
 The PALMS instrument showed differences in the proportions of the four classes of particles within the layer (Figure 3) compared to the overall statistics for the selected ITCT data set (Table 2). Within the CO layer, the fraction of particles containing NO+ along with sulfate and carbonaceous matter increased relative to the data set average and to the regions above and below the layer. This class of particle composition is typical of mixed sulfate-organic particles with nitrate and/or ammonium associated with urban and industrial emissions [Lee et al., 2002]. However, the frequency statistics within this CO-rich layer must be evaluated with consideration for the fact that most of the particles within the layer were smaller than 0.2 μm, the smallest size that the PALMS instrument can detect.
 In urban/industrial plumes transported over the western Pacific Ocean from Asia sampled during the ACE-Asia project, most of the volatile mass of submicron particles was composed of sulfate-ammonium-organic compounds [Bahreini et al., 2003]. Such particles should be hygroscopic, and would likely be removed by precipitation scavenging during passage through the WCBs as detailed by Cooper et al. . The high number concentrations of sub-0.2 μm particles we sampled near the California coast were probably the result of new particle formation and subsequent condensation by organic and inorganic gases oxidized following passage through the cyclonic systems. Because the volume (mass) concentration of these particles was very small, they could have originated from condensation of very small amounts of precursor gases (e.g., <40 pptv of SO2). Water soluble ionic components and particle volume were not detectably enhanced above background in this layer. In contrast, in a similar situation encountered on 17 May 2003 and discussed in section 3.3 below, mixing ratios of SO2 exceeding 1 ppbv remained following precipitation scavenging, allowing substantial secondary mass growth of particles.
3.2.2. 5 May: Characteristics of Layers Rich in Coarse Particles
 The volume of coarse particles (those with Dp > 1.0 μm) was strongly enhanced below the CO layer (Figures 2d and 2e). The PALMS single particle mass spectrometer, which detected particles with Dp from 0.2 to 5 μm, provided statistical information on particle composition for the coarse particles, though it sampled particles with Dp > 3 μm with <50% efficiency. Within the layer enhanced in coarse particles, the PALMS showed a substantial increase in the fraction of particles containing crustal material (Figure 3), and decreased fractions of particles in the carbonaceous-sulfate and carbonaceous-sulfate-NO+ classes, compared with the data set averages (Table 2). The layer rich in coarse particles did not have substantial enhancements in NOy or CO or other relatively long-lived anthropogenic tracers such as ethane and propane.
 It is conceivable that the crustal particles could have been transported along with the gas-phase pollutants and then fallen by gravitational settling, leaving the observed vertical structure of a layer rich in gas-phase pollutants overlying a layer of coarse dust particles. However, the following evidence suggests that the crustal particles did not sediment from the layer of gas-phase pollution above: (1) The fall velocities of spherical particles with densities of 2 × 103 kg m−3 (2 g cm−3) and diameters of 1, 3 and 5 μm are ∼7, 50, and 150 m day−1, respectively, at the measurement pressure of 600 hPa and temperature of 263 K. If transport from Asian sources took ∼7 days [Cooper et al., 2004], the largest particles would have fallen more than 1 km from their original altitude. However, steady sedimentation over a transport time of several days would have produced a gradient in the particle size distribution, with the largest particles having fallen to the lowest altitudes. This size sorting with altitude is not observed; instead, there is an abrupt discontinuity in coarse particle characteristics at the lower edge of the CO-rich layer (Figure 2d). (2) There are gradients in wind velocity and relative humidity between the CO layer and the coarse particle layer (Figure 2f), suggesting that differential horizontal advection, rather than particle sedimentation from a relatively homogeneous airmass, led to the observed vertical structure. (3) A layer of coarse particles was seen again at altitudes near 4 km at 22:15 UTC during a profile (Figure 1). In this case, CO concentrations above the coarse particle layer did not exceed 150 ppbv. Thus layers of soil particles were seen both with and without CO-rich layers above, suggesting there was no direct linkage between the two types of layers.
 To help identify possible sources for the crustal particles, 8-day backward trajectories were calculated from locations along the flight track within the layer. Most of the trajectories (not shown) circled the Pacific Ocean anticyclonically for the duration of the 8-day calculation and indicated slow descent. These trajectories contrast with those calculated for the CO layer above, which indicated relatively rapid transport from Asian sources [Cooper et al., 2004; Nowak et al., 2004]. The lack of obvious sources for the layer of coarse, crustal particles suggests either that the particles were lofted much more than 8 days prior to detection, or that the trajectory simulations are grossly inaccurate. In either case, we cannot readily identify the source of the particles.
3.2.3. 5 May: Characteristics of Layers Rich in Acetonitrile
 A third layer with different gas-phase and particle characteristics, this one above the concentrated CO layer from 5.3 to 6.8 km altitude, was also present on 5 May (Figures 2 and 3). Mixing ratios of acetonitrile (CH3CN) and methyl chloride (CH3Cl) were enhanced at altitudes above 6.8 km [de Gouw et al., 2004]. Acetonitrile and methyl chloride are tracers of biomass and/or biofuel combustion [de Gouw et al., 2003b]. Short-chain alkanes, which are indicators of anthropogenic pollution sources, were not enhanced within this upper, acetonitrile-rich layer, although they were quite strongly enhanced in the CO-rich layer below [de Gouw et al., 2004]. The mixing ratio of CO was modestly enhanced within the upper, acetonitrile-rich layer, reaching a peak value of 197 ppbv.
 While not clearly evident in Figure 2e due to scale, the volume of particles with Dp from 0.15–1.0 μm slightly increased, from a mean of 0.97 μm3cm−3 within the anthropogenic layer between 5.3 to 6.8 km altitude, to a mean of 1.25 μm3cm−3 above 7.2 km. The fraction of particles containing potassium was substantially higher within the acetonitrile-rich layer compared with the lower layers (Figure 3) or with the mission average (Table 2). The presence of potassium in the PALMS particle spectra (in the absence of sea-salt constituents) is one of the key indicators used to classify particles as originating from biomass burning sources [Hudson et al., 2004]. The measured mass concentration of water-soluble, submicron K+ did not increase significantly, indicating either that the potassium detected by the PALMS instrument was a minor fraction of the total particle mass, or that the potassium was not water soluble.
 The observed particle and trace gas characteristics and meteorological analyses indicate that this upper, acetonitrile-rich layer was transported from unspecified sources of biomass combustion in Asia [de Gouw et al., 2004]. The CO-rich layer between 5.3 and 6.8 km altitude originated from sources in eastern Asia and was transported through midlatitude cyclones to the point of measurement [Cooper et al., 2004; de Gouw et al., 2004; Nowak et al., 2004]. We have shown that below the CO layer lay a region containing enhanced concentrations of coarse, crustal particles with little anthropogenic signature and no obvious direct source. Thus in one vertical profile, three distinctly different aerosol layers were found to contain diverse trace gas and particle properties, indicative of different sources for each layer. These layers were vertically discrete and showed evidence of vertically decoupled, long-range transport from different sources [de Gouw et al., 2004].
3.3. 17 May 2002
 In this section, a second significant aerosol transport event encountered by the WP-3D during the ITCT 2002 project is examined in detail. On 17 May 2002 the aircraft flew a northeast-southwest flight track west of California approximately perpendicular to airflow that was predicted by tracer transport models to contain emissions from Asia. The aircraft encountered a layer of enhanced CO, NOy, HNO3, O3, CO2, H2SO4, SO2, C2H6, and other gas-phase indicators of pollution ∼270 km west of central California between 2.1 and 3.4 km above the surface (Figures 4 and 5). The same or a very similar layer was also observed at approximately the same equivalent potential temperature [Nowak et al., 2004] between 1.8 and 2.7 km during the final descent of the flight just west of Monterey, California, and between 3.1 and 4.1 km ∼650 and ∼850 km west of southern California (Figures 4 and 5).
3.3.1. 17 May: Characteristics of the Layer
 The layer observed on 17 May was unique in the analyzed data set for the following combination of measurements (Figure 5): (1) mixing ratios of SO2 reached ∼600 pptv (not shown), significantly exceeding the detection limit of 350 pptv; (2) high concentrations of H2SO4 (∼2 × 107 molecules cm−3) were observed despite the presence of substantial particle surface concentrations of ∼100 μm2cm−3 (not shown), which represents a large sink for H2SO4; (3) NOy was substantially elevated and >90% of the observed enhancement was due to HNO3; (4) the highest observed concentration of submicron particulate sulfate in the analyzed data set, 3.05 μg m−3, was within this layer; (5) O3 (not shown) was substantially enhanced above background mixing ratios and positively correlated with NOy [Nowak et al., 2004]; (6) CO reached values exceeding 240 ppbv; and (7) the relative humidity of the layer (not shown) was <2%.
 Within the most concentrated portion of the pollutant-rich layer, particle volume exceeded 4 μm3cm−3; >90% of this mass was associated with particles in a single accumulation mode with a mass median Dp of 0.3 μm, while the remaining volume was associated with coarse particles (Figure 5d). Particle number concentrations were ∼800 cm−3, and were dominated by a mode with a number median Dp of 0.14 μm (Figure 5c). Unlike the cases encountered on 5 May, there was no evidence of recent new particle formation within this layer. Further, while coarse particle concentrations were slightly enhanced within the polluted layer on 17 May (Figure 5d), coarse particles did not dominate the mass as they did in the crustal particle-rich layers encountered on 5 May.
 From the positive mass spectra recorded by the PALMS instrument, 13% of the analyzed particles were determined to be crustal, about 2 times the median for the data set (Table 2). Particles containing potassium represented 16% of the particles, while the sulfate-carbonaceous and sulfate-carbonaceous-NO+ classes comprised 35% and 22% of the total analyzed particles. Compared with the data set averages, these proportions indicate relatively fewer of the potassium-containing particles, a near-average proportion of sulfate-carbonaceous-NO+ particles, and a larger fraction of sulfate-carbonaceous particles. The PALMS instrument is not highly sensitive to the positive ions produced from particulate sulfate. However, in both positive and negative spectra from the PALMS an unusually high fraction of the ion current from these particles was associated with sulfate mass peaks, with relatively few carbonaceous and NO+ ions present. The observed spectra were qualitatively similar to those obtained from stratospheric sulfate particles, suggesting a predominantly sulfuric acid composition. The negative spectra did not show the large NO2−, NO3−, and oxygenated organic peaks typical of secondary particles in urban environments [Lee et al., 2003]. At CO mixing ratios exceeding 200 ppbv, >85% of all the particles analyzed for negative ion composition by the PALMS instrument had sulfate-related ions contributing >80% of the total negative ion signal. Ions from alkali metals (which the PALMS is especially sensitive to), particularly Li+, were also detected on a small fraction of the particles.
 Particle volume and the submicron sulfate concentrations measured by the PILS were highly correlated in the vicinity of the four encounters with the layer in which PILS samples were taken (Figure 6). An non-zero intercept indicates that a small percentage of the particle volume within the layers was not associated with sulfate compounds. The expected relationship between submicron sulfate and submicron volume can be determined if the composition of the particles is known. The PILS detected no enhancements above background in the concentrations of bulk submicron water soluble inorganic ions other than sulfate. Given the similarity of the PALMS spectra for the particles within these CO-rich layers and the stratospheric sulfate aerosol, a composition of sulfuric acid in water is assumed. Using the Aerosol Inorganics Model II (AIM II; http://mae.ucdavis.edu/wexler/aim.htm) [Clegg et al., 1998], under the measured conditions (T ∼ 280 K, relative humidity 1–4%), a composition of 70–75 weight% H2SO4 and density of 1.6 g cm−3 is calculated. The relationship between particle volume and sulfate for this composition is consistent with the measurements (Figure 6). A composition of (NH4)2SO4 is also consistent with the measurements (Figure 6). However, the measured concentrations of H2SO4(g), which are within the range of modeled equilibrium vapor pressures for H2SO4/H2O at these humidities and temperatures, as well as the PALMS mass spectra, strongly suggest that the particles are not substantially neutralized.
3.3.2. 17 May: Transport of the Layer
 Multiple 8-day backward trajectories were calculated using gridded three-dimensional meteorological fields [Cooper et al., 2004]. The output from trajectory simulations (Figure 7), initialized in a 45 × 45 × 0.7 km grid with 5 × 5 × 0.1 km resolution centered at the location of the observed maximum in pollutant concentrations at 22:40 UTC (Figures 4 and 5), shows that the parcels were transported in slowly descending air throughout the 8-day period. At the earliest time of the calculation, 8 days prior to measurement, most of the parcels were at altitudes from ∼6–7 km within the warm sector of a midlatitude wave cyclone (Figures 7b and 7d); relative humidities for these parcels were at or near 100%.
 To investigate potential sources of the pollution, the parcel locations were calculated further back in time, to 14 days prior to the encounter with the layer. These additional trajectory simulations indicate that many of the parcels were transported from altitudes as low as 1.2 km over eastern Asia (Figure 7c) prior to ascent within the Pacific cyclone. During this ascent to near 7 km, this presumably polluted air encountered substantial cloud formation and precipitation. Clearly, this interpretation must be tempered by the large uncertainties associated with the long transport times in the trajectory calculations [Cooper et al., 2004]; however, it is consistent with the observed low relative humidity and enhanced concentrations of indicators of pollution, such as CO, O3, NOy, alkanes, and acetylene, within the layer [de Gouw et al., 2004, 2003b; Nowak et al., 2004].
 The trajectories imply that the pollutant-rich layer was transported upward through a midlatitude cyclone followed by gradual descent over a period of ∼8 days prior to reaching the point of measurement near the California coast. The processes active during this transport are shown schematically in Figure 8. Incorporation of highly soluble gas-phase species, such as HNO3, H2SO4, and NH3, into cloud droplets and subsequent removal of these compounds and their reaction products by precipitation is expected during transport within WCBs within midlatitude cyclones [Crutzen and Lawrence, 2000; Miyazaki et al., 2003]. Hygroscopic particles, including those of mixed sulfate-nitrate-organic compositions common in polluted environments [Lee et al., 2003], are expected to nucleate or become incorporated in cloud droplets by diffusional and collisional attachment, and be effectively removed by precipitation. Insoluble gas-phase species such as PANs should be relatively unaffected by precipitation scavenging.
 The removal efficiency for modestly soluble and reactive species such as SO2 is more complex, and depends in part on the oxidation rate of SO2 with cloud droplets, which in turn depends upon the pH of the cloud droplets and the presence of oxidizing agents such as peroxides, dissolved metal ions, and O3. Most numerical simulations of transport in cloud systems suggest that substantial fractions of SO2 can be transported even through vigorous liquid convective clouds [Barth, 1994; Crutzen and Lawrence, 2000; Kreidenweis et al., 1997]. Once cloud droplets freeze, the uptake of most species effectively ends and gas-phase scavenging becomes less significant [Crutzen and Lawrence, 2000], although some heterogeneous reactions may continue until ice surfaces are passivated [Clegg and Abbat, 2001]. Thus if a highly polluted parcel of air travels through the precipitation formation region of an active, midlatitude, springtime cyclone, one might expect to see modest mixing ratios of SO2, relatively high concentrations of PANs and other insoluble gases [Miyazaki et al., 2003], and low number densities of hygroscopic particles or of soluble gases such as H2SO4, HNO3, or NH3, immediately downwind of the system. During subsequent, gradual descent, PAN would thermally decompose and oxidize to HNO3, while the SO2 would oxidize to form H2SO4. This H2SO4 would nucleate new particles and rapidly condense to form particulate mass. For the estimated parcel conditions (Table 3), >80% of transported PAN would have thermally decomposed [Roberts and Bertman, 1992], and the NO2 produced would have oxidized within ∼1 day to produce HNO3 at typical free tropospheric OH concentrations [Seinfeld and Pandis, 1998]. Thus this conceptual model of frontal uplift, scavenging, and transport (Figure 8) is qualitatively consistent with the HNO3, SO2, H2SO4, and particle composition observed in the polluted layer on 17 May and with observations of NOy partitioning in the outflow of frontal systems in the western Pacific Ocean downwind of Asia [Miyazaki et al., 2003].
Table 3. Parameters Used in Model Simulation of Particle Nucleation and Growth and Comparable Measurements at 22:40 UTC on 17 May 2003a
Ion Source, cm−3s−1
Model N, cm−3
Model S, μm2 cm−3
SO4=, μg m−3
SO4=, μg m−3
N and S are, respectively, number and surface concentrations at STP. Initial time for the model was 00 UTC on 10 May 2003. Observations (at 8 days) are 90th percentile values within the polluted layer, except for SO4=, which is for the sample at the highest CO mixing ratio. Uncertainties represent possible measurement biases (Table 1).
NA, not applicable.
OH was prescribed to a noontime maximum of 3 × 106cm−3 throughout the simulation, and initial SO2 to 1.3 ppbv, to produce these values after 8 days of simulation.
3.3.3. 17 May: Simulation of Particle Nucleation and Growth
 We evaluate here whether the observed concentrations of SO2, H2SO4, particulate sulfate, and particle number, surface area, and volume are consistent with the transport scenario outlined in section 3.3.2 (Figure 8). This evaluation is accomplished by numerically modeling the nucleation and growth of particles in a hypothetical air parcel that exited the cloudy region of the wave cyclone and was then transported over an 8-day period to the point of observation. The calculated size distribution is compared with the observed size distribution to evaluate whether the gross characteristics of the measurement, such as the presence of a single mode with peak Dp of ∼0.14 μm, and sulfate mass, and particle number and surface concentrations given in Table 3, can be explained by the conditions and processes within the parcel model.
 A numerical model of particle nucleation and growth in the H2SO4-H2O system combined with a sectional particle model [Lovejoy et al., 2004] was used to simulate the formation and growth of particles. This box model simulates the formation of new particles by ion-assisted binary nucleation. After stable neutral particles are formed, they grow by condensation and coagulation [Raes and Janssens, 1986].
 The model was run for the approximate parcel conditions as calculated from the trajectory simulations beginning from the time the parcel began its descent from the cloudy conditions. The conditions for the base case simulation (Table 3) were established as follows: (1) The parcel was assumed to exit the cloud system, 8 days prior to measurement, at an altitude of ∼7 km. The parcel maintained this altitude for 3 days, then descended to 4 km, where it remained for 4 days. The parcel then descended to 3 km and remained there until the time of measurement. (2) Initial H2SO4 and particle surface and number concentrations were presumed to be zero, due to effective precipitation scavenging. (3) The initial SO2 mixing ratio, 1.3 ppbv, was calculated from the total sulfur present in particles and gases at the time of observation. (4) OH concentrations, which control the H2SO4 source strength via the oxidation of SO2, were prescribed using a sinusoidal diurnal cycle to give a peak noontime OH concentration of 3 × 106 cm−3, sufficient to oxidize the 1.3 ppbv of SO2 assumed to be initially present to produce the observed particulate sulfate. These prescribed OH concentrations were compared against a photochemical box model [Chen et al., 2001], which produced a similar noontime maximum OH concentration of 2 × 106 cm−3. (5) Source strengths of gaseous ions [Arnold, 1980; Rosen et al., 1985] were prescribed to vary from 7–30 ions cm−3 with altitude (Table 3). (6) The coarse fraction of the particles, which contributed ∼8% to the total observed particle volume and ∼1% to the total surface area at the time of observation, was ignored throughout the calculation.
 The results of the numerical simulation (Table 3, Figure 9) approximate the observed particle number, surface area, and submicron particulate sulfate concentrations at the conclusion of the 8-day run. At this time, the concentration of H2SO4 of 2.7 × 107 cm−3 calculated by the simulation for 2.5 hrs after local solar noon was similar to the observations of 2.0(±0.7) × 107 cm−3 measured by the CIMS instrument. The model produced a unimodal particle size distribution centered at a Dp near 0.13 μm, similar to the measured distribution centered at 0.14 μm.
 At the conditions of the parcel exiting the cloud system 8 days prior to measurement, with a high production rate of H2SO4, low temperature, low preexisting particle surface area, and high relative humidity, the nucleation of new H2SO4-H2O particles was likely even in the absence of ions to assist in the process. Sensitivity tests (Figure 9) showed that the final size and mass of the particles depended mostly on the H2SO4 production rate and the presence or absence of preexisting particle surfaces.
 This analysis indicates that the observed H2SO4 concentrations and particle characteristics within the polluted layer can be explained by transport of SO2 through the cyclonic system following advection from Asian sources, and that the particle mass was mostly secondary in nature. Owing to the high production rate of H2SO4 (constrained in the model by the observed SO2, H2SO4 and particle surface concentrations), the assumed absence of preexisting particle surface area at the start of the simulation, and favorable thermodynamic conditions, the nucleated particles grew rapidly, reaching sizes effective as cloud condensation nuclei within 1 day after nucleation (Figure 9).
 The background, springtime, free tropospheric aerosol over the eastern Pacific Ocean was found to have low concentrations of particle surface area, volume, and inorganic ionic species, and particle number concentrations of several hundred per cm−3. Approximately equal number fractions of particles with Dp from 0.2 to 5.0 μm were composed of sulfate with carbonaceous material, sulfate with carbonaceous material and NO+, and potassium-containing material, while particles containing crustal elements were less frequent.
 Superimposed on this background, four distinct classes of aerosols transported from terrestrial sources were encountered: (1) high mass concentrations of soil particles not associated with anthropogenic pollutants, (2) particles from biomass/biofuel burning sources at modest number and mass concentrations, (3) high number concentrations of sub-0.1 μm secondary particles with insignificant mass, associated with substantial gas-phase pollution (but no detectable sulfurous gases), and (4) high number and mass concentrations of secondary, accumulation-mode sulfuric acid particles associated with gas-phase transport of SO2 and in situ production of H2SO4. Aerosols composed of mixed crustal and anthropogenic particles combined with gas-phase tracers of anthropogenic activity, as have been reported previously [Clarke et al., 2001; Jaffe et al., 2003], were not found in the few transport events that were encountered during the airborne portion of the ITCT field program.
 The transported aerosols were found in layers with vertical extents of 0.3 to 1.5 km, but with maximum horizontal dimensions of at least 950 km. In any given event, these layers were generally restricted to a narrow range of altitudes, and appeared to have been separated from surrounding air by vertical shear in the horizontal wind. The thermodynamic and wind profiles imply that the layers were formed by differential advection with minimal mixing with the surrounding airmasses. In the cases examined, the particle layers associated with gas-phase indicators of pollution were found well above the MBL, and could be traced to cyclonic systems over the Pacific Ocean. Thus the conceptual model of lifting of polluted air from the boundary layer due to cyclonic activity, possibly in the WCB of the cyclone [Cooper et al., 2002; Liu et al., 2003; Miyazaki et al., 2003], followed by anticyclonic, quasi-isentropic descent in a stable, sheared layer, is generally consistent with the observed composition and structure of the polluted layers and with the backtrajectory analyses.
 A complete explanation of the microphysical and chemical variability of the aerosol requires consideration of sources as well as atmospheric processing during transport. In the case of 5 May, when particulate sulfate and sulfurous gases were not detected within a layer of CO with mixing ratios exceeding 250 ppbv, SO2 may have already been converted to aerosol sulfate prior to incorporation of the pollution into the cyclone and lifting to the free troposphere. In such a case, precipitation scavenging would be expected to remove the sulfate mass, resulting in long-range transport of only small quantities of gas-phase particle precursors. Alternatively, the source region may have had low SO2 emissions, characteristic of a modern, urban source [Streets and Waldhoff, 2000]. The condensable gases transported were evidently sufficient to nucleate and grow some particles to Dp < 0.1 μm (Figure 2c), but they were insufficient for substantial mass production (Figure 2d). On 17 May, when SO2, H2SO4, particulate sulfate, and HNO3 were present in quite large quantities within a polluted layer, SO2 from a source region was likely carried through the cyclonic system. A substantial portion of this SO2 remained after scavenging of most of the anthropogenic particles and soluble gases within the cyclonic system. As indicated by the numerical modeling of particle formation and growth, this SO2 subsequently oxidized to H2SO4(g), formed new particles, and further condensed to form substantial secondary particulate H2SO4 mass during the slow, descending transport downstream of the cyclone. Similarly, NOy was probably transported through the precipitation region as insoluble PAN, and subsequently descended, thermally decomposed, and the resultant NO2 was oxidized by OH to produce the observed HNO3.
 For the case of the layer rich in acetonitrile and particles associated with biomass combustion measured on 5 May, trajectories indicate that unspecified regions in eastern Asia were the most likely source for the observed particles and gases [de Gouw et al., 2004]. In this case, only slight enhancements in particle volume were detected and water-soluble inorganic species were not detectably enhanced in the layers. These findings suggest that only a few, possibly non-hygroscopic, particles survived transport from the source regions, and that SO2 and PANs were not present in substantial quantities. For the layer rich in crustal particles encountered below the CO layer on 5 May, trajectories do not indicate any obvious source of the particles. It is likely that these soil particles were lifted during high wind events and carried with little modification into the free troposphere without transport through a cloud system.
 Observations and analysis from recent field studies over the western Pacific Ocean near Japan indicate that most springtime transport from the polluted Asian continental boundary layer to the free troposphere is associated with convective and WCB transport in cyclonic systems [Miyazaki et al., 2003]. This evidence, and the variety of aerosol properties reported here, suggest that sophisticated parameterizations regarding sources, transport, cloud processing, and gas-phase and liquid-phase chemistry involving particle precursors are required to accurately represent particle transport and evolution in chemical transport models. Such models may also need to include relatively sophisticated parameterizations of particle nucleation processes and condensational and coagulational growth to realistically simulate the particle properties following long-range transport.
 The presence of substantial concentrations of nearly pure H2SO4 particles in layers of considerable horizontal and vertical extent in the lower troposphere approximately two weeks removed from potential sources is surprising, and the chemical, optical, and cloud-forming effects of such particles may be considerable. For example, a variety of multiphase chemical reactions have been associated with tropospheric sulfuric acid droplets [Ravishankara, 1997]. Additionally, the light extinction due to sulfuric acid particles is greatly enhanced compared to most other particle compositions due to the propensity of sulfuric acid particles to grow by water uptake even at modest relative humidities [Tang, 1996]. This hygroscopicity suggests a larger potential climatic effect if particles are transported intercontinental distances in the form of sulfuric acid droplets rather than as fully or partially neutralized sulfates.
 Many thanks to M. Trainer for discussions and manuscript suggestions. GMS-5 satellite images were provided by the Space Science and Engineering Center, University of Wisconsin-Madison. GOES-10 satellite images were provided by UNIDATA Internet delivery and displayed using McIDAS software. This work was supported in part by NOAA's Office of Global Programs under grants NA06GP0410, NA16GP1478, and NA16GP1579, and by the National Science Foundation. J. B. Nowak was supported by a National Research Council Research Associateship Award at the NOAA Aeronomy Laboratory.